(Last updated on 2020-04-20)
We have learned how South Korea has implemented a trace-test-isolate strategy - universally across the country and consistently from the outset of the epidemic. Then, how does South Korean flattened curve look like, compared to other countries? This report compares metrics of the epidemic curve in South Korea and other high-resource settings, specifically OECD member countries, focusing on the confirmed cases. In addition, Taiwan and Singapore are included considering their exemplary handling of the epidemic.
Hover over each figure to see values and more options.
See data sources and methods at the end.
Before we check the overall epidemic curve to date, let’s briefly compared cumulative incidence, case fatality, and mortality rates. Cumulative incidence rate: number of confirmed cases per 100,000 population. It may not be a good indicator to compare, considering vastly different testing rates - especially in the beginning of the epidemic. Also, definitions of confirmed cases can be slightly different across countries. In South Korea, confirmed cases refer to laboratory confirmed cases.
Case fatality rate (CFR): number of deaths per 100 confirmed cases. CFR is sensitive to demographic and clinical characteristics of patients. CFR can be high in countries where testing has been limited to only people with clinical symptoms. In South Korea, patient population is relatively young, and there have been relatively more mild cases because of extensive tracing and testing.
Mortality rate: number of COVID-19 deaths per 100,000 population. Considering different testing rates and this may be the most appropriate indicator to compare impact of the epidemic across countries. Singapore, South Korea, and Taiwan - three countries with relatively similar strategies - all have substantially low number of COVID-19 deaths per population.
The epidemic curve was constructed based on the numebr of new confirmed cases each day. Considering vastly different population sizes, the number of new confirmed cases each day per 100,000 population was used for comparison purposes, instead of the absolute number. Finally, 5-day rolling averages were used to avoid any isolated peaks/drop, which can be caused by various reasons other than the true course of epidemic itself (e.g., changes in definition, and lab process delay). See annex for further information about methods.
Let’s focus on the first wave of the epidemic, which we define it starts when cumulative incidence exceeded 1 confirmed case per 100,000 population. There are three distinctive phases:
With successful control of the epidemic, we will see both narrow and also short peak.
The data (i.e., smoothed number of new confirmed cases each day per 100,000 population) clearly show the three phases in South Korea. The cumulative incidence rate exceeded 1 per 100,000 population on 2020-02-23, and the daily new cases reached the peak (1.2 new confirmed cases per 100,000 population) on 2020-03-01. The country has been well into the relative stable phase since 2020-03-11, for 39 days.
Several other countries also have entered the relative stable phase, although the first wave started later. Holding beginning of the first wave on day 0, the following figure compares the length and height of the peak. Compared to South Korea, the height of the peak tends to be slightly higher - except in Greece. Also, a few countries have just entered the relative stable phase, and it is important to keep the number of new infections low without any rebound.
Meanwhile, a majority of countries have not yet entered the relative stable phase. Most of these countries have substantially high peak, compared to the above countries. Meanwhile, a few patterns emerge.
Some would reach the stable phase soon if the pace continues (e.g., Austria, Iceland, Luxembourg, and Switzerland).
Some have had multiple rebounds. At each time, the number of new infection decreased substantially but then increased again (e.g., Estonia, France, Norway, and Slovakia).
Some passed the peak only recently (e.g., Belgium, Israel, Sweden, United Kingdom, and United States). In these countries, the peak can move up since fluctuating increases were observed in other countries, as noted above.
Some have entered the first wave, and the curve is going up slowly (e.g., Japan and Mexico). The level is low, but also there is no sign of decrease.
These countries are presented in two figures, due to varying range of peak height. Top panel has countries with relatively lower peak, and the bottom panel shows countries with higher peak.
Still, Taiwan has shown the best results of epidemic control - with very slow or no progression of the first wave. Singapore had controlled the epidemic successfully for over two months, but the new infections has increased rapidly recently.
To summarize, below first shows the timeline of the first wave and its phases. South Korea had the first wave earlier than most other OECD countries. It also has one of the shortest length of the peak. Again, the peak length is still to be determined in many countries (with no yellow dot below).
The following compares the height of the peak (the maximum number of new confirmed cases per 100,000 population) by country.
And, finally, if we put together the height of the peak (on a log scale, Y axis) and time to the peak (distance between the red and orange dots above)…
METHODS
Data:
1. All COVID-19 data (i.e., cumulative confirmed cases and deaths by day) come from JHU/CSSE. Accessed on 2020-04-20.
2. All data on country population come from UN World Population Prospects 2019 Revision. Accessed on April 18, 2020.
Measures:
The number of new confirmed cases on each date was calculated based on the difference between cumulative numbers over two days. Then, a 7-day rolling average was calculated for, hereinafter referred to as the smoothed number of new confirmed cases. Then, the smoothed number was divided by the total population in the country, as the smoothed number of new confiremd cases per 100,000 popualtion.
* The first wave was defined to start when the cumulative incidence rate exceeds 1 per 100,000 population. The smoothed number of new confirmed cases at the start of the first wave varies by country, but it averages around 0.5 per 100,000.
* A peak date of the wave was the date when the smoothed number of new confiremd cases per 100,000 popualtion was the maximum.
* A date entering the stable phase was when the smoothed number of new confiremd cases per 100,000 popualtion was less than the number on the start date of the wave.
COVID-mortality (either CFR or preferably COVID-specific mortality rate in the population) may be more comparable indicator, given considerably different testing strategies and testing rates. However, countries are currently at different stages of the curve, and comparison of mortality data would be more appropriate once most countries are in a similar phase of the epidemic (i.e., well in to the stable phase).
See GitHub for data, code, and more information. For typos, errors, and questions, contact me at www.isquared.global
Making Data Delicious, One Byte at a Time, in good times and bad times.